Enzymes catalyze the reactions essential to biology. In broad strokes, researchers understand how they work, but not how they work well. The complexity of enzymes makes it hard to identify which parts are important to catalysis. The amino acids in the reaction site are essential, but not enough to create highly active enzymes. Distant residues are also important and some residues are helpful only in some cases, when other residues are also present (epistasis). Identifying all residues that contribute to catalysis s an essential first step to learn why enzymes work well. To identify these residues, we propose to recreate ancestral enzymes. This recreation reduces complexity because it focuses on those few mutations directly associated with historical functional changes, thereby avoiding the pitfalls inherent to using (incomplete) knowledge based approaches. The number of differences that need to be screened is reduced from hundreds to tens enabling the systematic use of site directed mutagenesis to explore the contribution that each mutation makes to changes in function. Ancestral sequence reconstruction also has the potential of identify epistatic sites that form local optima on which directed evolution methods get trapped. Parallel acquisitions and reversals in function within superfamilies allow alternative solutions to be explored. Preliminary data show that the approach is feasible: ancestral enzymes have been synthesized, shown to be active and more promiscuous than modern specialist descendants and sites critical to high activity outside active sites have been found. Recreating catalytically promiscuous ancestral stem cell enzymes allows us to recapture the functional plasticity needed to identify what determines different types of activity in the same protein fold.